@misc{indiciae9ca33b40bd28,
  title = {How Transformers Learn to Plan via Multi-Token Prediction},
  author = {Jianhao Huang and Zhanpeng Zhou and Renqiu Xia and Baharan Mirzasoleiman and Weijie Su and Wei Huang},
  year = {2026},
  url = {https://arxiv.org/abs/2604.11912},
  note = {Source identifier: 2604.11912}
}